StrictLLM

StrictLLM is a secure chat interface for working with large language models. It supports coordinating multiple models in one workspace, creating personas with custom system prompts and assigned LLM providers, and invoking personas via @mentions. Tool use is governed by Always Allow, Always Ask, or Always Deny permissions per persona, and MCP servers can be connected and optionally run in Podman containers with configurable network access. Pricing tiers are Starter ($29/month), Pro ($79/month), and Enterprise (contact sales). The site states credentials are stored with platform-native encryption and the app can run offline with Pro or Enterprise licenses, with a 7-day offline grace period on Starter.

Key features

• Multi-LLM conversations in one workspace
• Persona mentions to hand work to specialist personas
• Tool-level permissions (Always Allow, Always Ask, Always Deny)
• Persona management with custom system prompts
• MCP server integration for private data and custom tools
• MCP server sandboxing in Podman containers with configurable network access
• Assignment of different LLM providers to different personas
• Support for OpenAI, Anthropic, OpenRouter, and local models via LM Studio, llama.cpp, Ollama, and OpenAI-compatible endpoints
• Offline mode for Pro and Enterprise licenses
• Audit logs (Enterprise)
• Single-Sign On support (Enterprise)

Best suited for

• Individuals wanting a secure multi-model chat client
• Power users orchestrating multiple AI personas
• Teams connecting AI to private data via MCP servers
• Organizations requiring audit logs and SSO

Pricing

Starter $29/month (3 personas, no orchestration); Pro $79/month (unlimited personas, full orchestration, offline mode, mobile access); Enterprise: contact us